Correct sampling in observational studies.

Introduction: Teaching evaluation by university students is one of essential and controversial processes in higher education. The core of debates over student evaluation of professors is the possibility of interference of factors irrelevant to professor rsquo;s teaching skills and capabilities. The main purpose of this research was to explore the confounding variables in teaching evaluation by students. Methods: This paper was a descriptive correlational study which employed structural equation modeling. The population included all BSc students of paramedics' faculties studying in the second semester of 20132014 academic year in Iran University of Medical Sciences. A number of 250 participants were selected through cluster sampling. The data collection tool was Ghazi Tabataei and Yousefi rsquo;s questionnaire modeled after Marks, Paswan, and Young with a five point Likert scale and 31 items. It measured six factors related to student evaluation of teaching. The data were analyzed by means of structural equation modeling technique. Results: A number of 215 questionnaires were analyzed. The findings showed that professorstudent interpersonal relationship, students' interest in the course, and professors' effort and involvement had a positive correlation with student evaluation, and course difficulty, dissatisfaction with the scores and the difficulty of assignments had a negative correlation with student evaluation of the professor. Conclusion: The results showed that confounding and sometimes confusing factors may affect student evaluation of teaching and cause false judgments or decisions.